Modeling Localness for Self-Attention Networks

TL;DR

This paper introduces a learnable Gaussian bias to model local context in self-attention networks, improving their ability to capture local dependencies while maintaining global context understanding.

Contribution

It proposes a novel localness modeling method as a learnable Gaussian bias integrated into self-attention, specifically applied to lower layers for better local context modeling.

Findings

Improves translation quality in Chinese-English and English-German tasks.

Enhances local dependency capturing without sacrificing global context.

Demonstrates universality across language pairs.

Abstract

Self-attention networks have proven to be of profound value for its strength of capturing global dependencies. In this work, we propose to model localness for self-attention networks, which enhances the ability of capturing useful local context. We cast localness modeling as a learnable Gaussian bias, which indicates the central and scope of the local region to be paid more attention. The bias is then incorporated into the original attention distribution to form a revised distribution. To maintain the strength of capturing long distance dependencies and enhance the ability of capturing short-range dependencies, we only apply localness modeling to lower layers of self-attention networks. Quantitative and qualitative analyses on Chinese-English and English-German translation tasks demonstrate the effectiveness and universality of the proposed approach.